15
$\begingroup$

This sounds a bit random, but it stems from a lecture in statistical genetics which I attended a while ago. We were shown a population lineage graph from which it was clear that most lineages eventually go extinct. The further you go in to the past, the smaller the percentage of the individuals living at that time whose lineage is still alive today. This probably makes sense in the context of evolution.

So I am curious about the speed of that process in human populations. I am aware that this speed, quite possibly, varies over time and geographically. Are there any estimates about that speed in European populations? To be more specific, I would take England at 1500 AD as a starting point. What is the percentage of the people living in England at that time whose lineage still exists?

EDIT

I am very grateful to @Richard Erickson for the reference to the Galton-Watson process. It is really fascinating that +70% of all known Chinese family names have become extinct. It would be really interesting to know more about the speed of the process of extinction - in particular in European populations.

Improve this question
$\endgroup$
8
  • 3
    $\begingroup$ It totally depends. This is difficult to answer based on just calculations because unless you have exact data for the population dynamics of the region. This is mostly an issue of bookkeeping (census). The further you go in the past the smaller was the percentage of the individuals living at that time whose lineage is still alive.: not necessarily true; imagine population bottlenecks. $\endgroup$
    – WYSIWYG
    Commented Aug 9, 2015 at 14:30
  • 1
    $\begingroup$ Assuming arguendo that a bottleneck killed all except 10 people in the population in 1500, right after your 1500 sample was taken. The percentage of people who still have their lineages extant today would be greatly reduced. $\endgroup$
    – March Ho
    Commented Aug 9, 2015 at 23:11
  • 2
    $\begingroup$ Interesting question +1. One can calculate such things using Coalescent Theory but the issue is that it assumes constant effective population size which is obviously very wrong for England. The Coalescence should have very long branches, much longer than expected from constant effective population size. I don't think there is an analytical method for dealing with growing population size. Also, the descendents went through several bottlenecks and rapid expansion so it definitely is not an easy case. I don't think that theory can help here, one will need to analyze data to answer this question. $\endgroup$
    – Remi.b
    Commented Aug 11, 2015 at 20:35
  • 6
    $\begingroup$ Check out Galton-Watson branching process models. They area closely related topic that focus on the probability of an individual surname going extinct. $\endgroup$
    – Richard Erickson
    Commented Aug 12, 2015 at 0:37
  • 2
    $\begingroup$ In reaction to @WYSIWYG's comment: The further you go in the past the smaller was the percentage of the individuals living at that time whose lineage is still alive is very likely to be true but not necessarily indeed. However, The further you go in the past the smaller (or equal) was the number of individuals living at that time whose lineage is still alive is always true. $\endgroup$
    – Remi.b
    Commented Aug 12, 2015 at 16:55

1 Answer 1

Reset to default
2
$\begingroup$

Here is a partial answer that also expands upon a previous comment:

Between 1900 and 2013, 20,000 English surnames had gone extinct according to the Daily Mail's summary of research on surnames. This is a partial answer because it only tells us how many names are gone and is from the wrong time period.

To completely answer your question, I suspect you need to conduct primary research. I would suggest the following broad steps:

  1. Gain access to the data needed to answer your question. British Census records only appear to available from 1841 to 1911 so you most likely need to sample old names (e.g., graveyard based studies).
  2. Use the data to parameterize a statistical model such as a Galton-Watson branching process models.

I doubt the answer to your question can quickly be found unless somebody has already done it. Although, if done well, it could probably be a masters' thesis. However, before embarking on such a project, I would suggest doing a good literature search to make sure it has not been done before.

Note: If anybody finds the answer, please feel free to take content from my answer and include it yours.

Improve this answer
$\endgroup$
4
  • $\begingroup$ Thanks. Daily Mails seems to refer (though not providing a direct link) to research done by ancestry.co.uk. I haven't checked that site yet but it would be another port of call. The thing is that family names are a sort of proxy variable. There was a lot of emigration from the UK. Also, people adjusted the spelling sometimes. More importantly, females traditionally took over the family name of their husband (so their own family name may become extinct while their lineage may continue). The emergence of double-barrelled family can also affect the stats. $\endgroup$
    – Nick
    Commented Jun 7, 2016 at 21:57
  • $\begingroup$ You're welcome. I know this data would be hard to come by. An academic researcher might be able to get it from ancestry.com or ancestry.co.uk. I know Google shares information with academics. Otherwise, it might also be machine learning exercise in reading in old church records or something like that. Good luck with the question though! $\endgroup$
    – Richard Erickson
    Commented Jun 7, 2016 at 22:10
  • 1
    $\begingroup$ I realize there it would be a lot easier if only a subgroup of the population is taken into account - the nobelty. I presume good records exist. The question is how to get them. $\endgroup$
    – Nick
    Commented Jun 10, 2016 at 19:23
  • $\begingroup$ @Nick, good point. Check out out Galton's work. He was afraid of the "good" aristocratic names going extinct. $\endgroup$
    – Richard Erickson
    Commented Jun 10, 2016 at 19:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .